Table of Contents

Overview

Capacity planning ensures your network can handle current workloads and scale for future growth. This guide covers bandwidth calculations, device capacity, performance monitoring, and growth planning.

Key Capacity Metrics

Bandwidth

Definition: Maximum data transfer rate

Measurement: bits per second (bps, Kbps, Mbps, Gbps)

Types:

  • Theoretical: Maximum possible (e.g., 1 Gbps Ethernet)
  • Actual: Real-world throughput (typically 70-95% of theoretical)
  • Available: Unused capacity at any moment

Throughput

Definition: Actual data successfully transferred

Factors affecting:

  • Protocol overhead (TCP/IP headers)
  • Errors and retransmissions
  • Congestion
  • Device processing capacity

Rule of Thumb: Expect 70-90% of rated bandwidth in real-world conditions

Latency

Definition: Time for packet to travel from source to destination

Measurement: milliseconds (ms)

Components:

  • Propagation Delay: Physical distance (light/electrical speed in medium)
  • Transmission Delay: Time to push bits onto wire
  • Processing Delay: Router/switch processing time
  • Queuing Delay: Waiting in buffers

Acceptable Latency:

  • Web browsing: < 100 ms
  • VoIP: < 150 ms (preferably < 30 ms)
  • Video conferencing: < 150 ms
  • Gaming: < 50 ms
  • Trading/Real-time: < 10 ms

Packet Loss

Definition: Percentage of packets that fail to reach destination

Acceptable Levels:

  • Data applications: < 1%
  • VoIP: < 0.5%
  • Video: < 0.1%

Jitter

Definition: Variation in packet arrival times

Impact: Critical for VoIP and video

Acceptable: < 30 ms for VoIP

Bandwidth Planning

User Bandwidth Requirements

Per-User Estimates (typical):

Activity Bandwidth Notes
Email 100-500 Kbps Text-based, occasional attachments
Web Browsing 1-5 Mbps Modern websites with images/video
VoIP Call 100 Kbps Per concurrent call
Video Conferencing 1-4 Mbps HD quality, per participant
Video Streaming 5-25 Mbps HD to 4K
File Downloads Varies Peaks based on file size
Cloud Applications 1-5 Mbps Office 365, Google Workspace
VDI/Remote Desktop 150 Kbps - 10 Mbps Depends on use case

Calculating Internet Bandwidth

Formula:

Total Bandwidth = (Users × Average per User) × Oversubscription Factor

Example: Small Office (20 users)

Assumptions:
- 20 users
- Average: 2 Mbps per user during work hours
- Oversubscription: 4:1 (not all users at max simultaneously)
- Growth: 25% over 3 years

Calculation:
Base: 20 users × 2 Mbps = 40 Mbps
With oversubscription: 40 Mbps ÷ 4 = 10 Mbps minimum
With growth: 10 Mbps × 1.25 = 12.5 Mbps

Recommendation: 25 Mbps circuit (room for peaks and growth)

Example: Medium Office (100 users)

Assumptions:
- 100 users
- Mix: 60% light (1 Mbps), 30% medium (3 Mbps), 10% heavy (10 Mbps)
- Oversubscription: 5:1
- Peak factor: 1.5× (spikes during lunch, start/end of day)

Calculation:
Light: 60 × 1 Mbps = 60 Mbps
Medium: 30 × 3 Mbps = 90 Mbps
Heavy: 10 × 10 Mbps = 100 Mbps
Total: 250 Mbps
With oversubscription: 250 ÷ 5 = 50 Mbps
With peak factor: 50 × 1.5 = 75 Mbps

Recommendation: 100 Mbps circuit minimum, 200 Mbps ideal

Oversubscription Ratios

Definition: Ratio of total possible bandwidth to available bandwidth

Common Ratios:

  • Access Layer: 20:1 to 40:1 (users to uplink)
  • Distribution Layer: 4:1 to 8:1
  • Core Layer: 1:1 to 2:1 (minimal oversubscription)
  • Internet: 4:1 to 10:1 (depends on usage patterns)

Example:

48-port gigabit switch with 10 Gbps uplink:
- Total capacity: 48 × 1 Gbps = 48 Gbps
- Uplink capacity: 10 Gbps
- Oversubscription: 48:10 or 4.8:1 ✅ (acceptable)

48-port gigabit switch with 1 Gbps uplink:
- Total capacity: 48 Gbps
- Uplink capacity: 1 Gbps
- Oversubscription: 48:1 ⚠️ (potential bottleneck)

Device Capacity

Switch Capacity

Factors:

Port Count: Number of devices supported

Switching Capacity: Internal bandwidth (measured in Gbps)

  • Formula: Port Count × Port Speed × 2 (full-duplex)
  • Example: 48-port gigabit = 48 × 1 × 2 = 96 Gbps

Forwarding Rate: Packets per second (pps)

  • Gigabit port: ~1.488 million pps
  • 10 Gbps port: ~14.88 million pps

Buffer Size: Packet queue capacity

  • Larger buffers = handle bursts better
  • Too large = increased latency

MAC Address Table: Maximum number of MAC addresses learned

  • Typical: 8,000-32,000 entries
  • Enterprise: 128,000+ entries

Router Capacity

Factors:

Throughput: Packet forwarding rate (Mbps/Gbps)

  • Often less than interface speed due to processing
  • Varies by features enabled (firewall, VPN, etc.)

Packets Per Second: Processing capacity

  • Small packets = more PPS required
  • 1 Gbps with 64-byte packets = 1.488 Mpps

Concurrent Sessions: Number of simultaneous connections

  • Residential: 10,000-50,000
  • SMB: 100,000-500,000
  • Enterprise: 1M-10M+

VPN Throughput: Encrypted traffic capacity

  • Usually 30-60% of routing throughput
  • Hardware acceleration improves performance

Wireless Access Point Capacity

Theoretical vs Real-World:

Standard Theoretical Real-World Clients/AP Notes
802.11n 600 Mbps 200-300 Mbps 25-30 2.4 GHz + 5 GHz
802.11ac 1.3 Gbps 500-800 Mbps 30-50 Wave 1
802.11ac 3.5 Gbps 1-1.5 Gbps 50-75 Wave 2
802.11ax (WiFi 6) 9.6 Gbps 2-4 Gbps 75-200 Better efficiency

Capacity Factors:

  • Client count: More clients = shared airtime
  • Client types: WiFi 6 clients more efficient
  • Channel width: Wider channels = more speed, less overlapping
  • Frequency: 5 GHz faster but shorter range than 2.4 GHz

Planning Guidelines:

  • Office: 1 AP per 2,000-3,000 sq ft, 30-50 users per AP
  • High-density (conference room): 1 AP per 1,000 sq ft, 50-75 users per AP
  • Warehouse: 1 AP per 10,000 sq ft, 10-20 users per AP

Server Capacity

Network Interface Requirements

1 Gbps NICs:

  • Small office file servers
  • Web servers (< 50 concurrent users)
  • Most virtual machines

10 Gbps NICs:

  • Medium enterprise file servers
  • Database servers
  • Virtualization hosts
  • High-traffic web servers

25/40/100 Gbps:

  • Data center storage
  • High-performance computing
  • Dense virtualization

Application-Specific

File Server:

Formula: (Users × Average Transfer Rate × Simultaneity Factor)

Example: 100 users
- Average transfer: 10 Mbps per user
- Simultaneity: 20% (20 users accessing simultaneously)

Calculation: 100 × 10 Mbps × 0.20 = 200 Mbps
Recommendation: 1 Gbps NIC (headroom for peaks)

Database Server:

Consider:
- Transaction rate (transactions/second)
- Average transaction size
- Query complexity
- Number of concurrent connections

Recommendation: 10 Gbps for > 100 concurrent users

Web Server:

Formula: (Concurrent Users × Page Size) / Page Load Time

Example:
- 500 concurrent users
- 2 MB average page size
- 2 second target load time

Calculation: (500 × 2 MB × 8 bits) / 2 sec = 4 Gbps
Recommendation: 10 Gbps NIC or load balancer across multiple 1 Gbps servers

Growth Planning

Forecasting Growth

Historical Analysis:

  1. Collect baseline data (6-12 months):

    • Bandwidth utilization
    • User count
    • Application usage
    • Peak periods

  2. Calculate growth rate:

    Monthly Growth % = ((Current - Previous) / Previous) × 100
Annual Growth % = Monthly Growth % × 12

  3. Project future needs:

    Future Capacity = Current Capacity × (1 + Growth Rate)^Years

Example:
Current: 100 Mbps
Growth: 15% annually
Planning: 3 years

Future = 100 × (1.15)^3 = 152 Mbps
Recommendation: 200 Mbps circuit (30% buffer)

Planning Horizons

Short-term (1 year):

  • Monitor current utilization
  • Address immediate bottlenecks
  • Plan for known changes (new hires, applications)

Medium-term (2-3 years):

  • Equipment refresh cycles
  • Technology upgrades (WiFi 6 → WiFi 7)
  • Capacity expansion (more switches, APs)

Long-term (5 years):

  • Architectural changes
  • New building/site additions
  • Major technology shifts (10 Gbps → 25 Gbps)

Capacity Triggers

When to Upgrade:

Internet Connection:

  • Sustained > 70% utilization during peak hours
  • Frequent complaints about slow access
  • New applications require bandwidth

Switches:

  • Ports exhausted (> 80% utilized)
  • Uplink saturation (> 70% sustained)
  • Switch CPU > 70%

Wireless:

  • Client count approaching AP maximum
  • RSSI < -70 dBm in coverage areas
  • Frequent disconnections or roaming issues

Servers:

  • NIC utilization > 70% sustained
  • Packet drops due to buffer overflows
  • Application response time degradation

Performance Monitoring

Key Performance Indicators (KPIs)

Network Availability:

Availability % = (Total Time - Downtime) / Total Time × 100

Target: 99.9% (43.8 minutes downtime/month)
Enterprise: 99.99% (4.38 minutes downtime/month)

Bandwidth Utilization:

Utilization % = (Current Traffic / Maximum Capacity) × 100

Healthy: < 70% peak
Warning: 70-85% peak
Critical: > 85% sustained

Packet Loss:

Loss % = (Packets Lost / Packets Sent) × 100

Target: < 0.1%

Latency (Ping):

Target:
- LAN: < 10 ms
- Internet: < 100 ms
- VoIP: < 150 ms

Monitoring Tools

Open Source:

  • PRTG (free tier): SNMP monitoring, bandwidth sensors
  • Nagios: Infrastructure monitoring, alerting
  • Zabbix: Comprehensive monitoring
  • LibreNMS: SNMP-based network monitoring
  • Cacti: Graphing, trending

Commercial:

  • SolarWinds Network Performance Monitor: Comprehensive, expensive
  • PRTG (paid): User-friendly, scalable
  • Datadog: Cloud-native, APM integration
  • Auvik: Cloud-managed, MSP-focused

Built-in:

  • Switch/router SNMP
  • NetFlow/sFlow
  • Syslog
  • RMON

Establishing Baselines

Process:

  1. Collect data for 2-4 weeks (capture all patterns)
  2. Identify patterns:
    • Peak usage times (morning login, lunch, end of day)
    • Day of week variations (Monday vs. Friday)
    • Monthly cycles (month-end reporting)
  3. Document normal ranges:
    • Average utilization
    • 95th percentile (billing standard)
    • Peak utilization
  4. Set alerting thresholds:
    • Warning: 70% of capacity
    • Critical: 85% of capacity

Capacity Planning Checklist

Assessment Phase

  • ☐ Document current users and devices
  • ☐ Measure current bandwidth utilization
  • ☐ Identify applications and their requirements
  • ☐ Review historical growth trends
  • ☐ Survey user satisfaction and complaints

Planning Phase

  • ☐ Calculate bandwidth requirements per user/application
  • ☐ Apply oversubscription ratios
  • ☐ Factor in growth projections (3-5 years)
  • ☐ Add buffer capacity (20-30%)
  • ☐ Consider redundancy requirements

Design Phase

  • ☐ Select appropriate circuit sizes
  • ☐ Design switch/router hierarchy with adequate uplinks
  • ☐ Plan WiFi AP density and placement
  • ☐ Size server NICs appropriately
  • ☐ Document expected performance metrics

Implementation Phase

  • ☐ Deploy monitoring tools
  • ☐ Establish performance baselines
  • ☐ Configure alerting thresholds
  • ☐ Document as-built capacity

Ongoing Phase

  • ☐ Monthly utilization reviews
  • ☐ Quarterly capacity reports
  • ☐ Annual growth assessment
  • ☐ Proactive upgrades before saturation

Real-World Examples

Example 1: Small Office Internet Upgrade

Scenario: 25-user office, frequent complaints about slow internet

Assessment:

  • Current: 50 Mbps cable internet
  • Measured peak: 48 Mbps (96% utilization)
  • Growth: 3 new hires planned this year

Analysis:

Current per-user: 50 Mbps / 25 = 2 Mbps
With new hires: 50 Mbps / 28 = 1.79 Mbps (insufficient)
Target per-user: 3 Mbps (comfortable)
Required: 28 × 3 Mbps = 84 Mbps
With buffer (25%): 84 × 1.25 = 105 Mbps

Solution: Upgrade to 100 Mbps (available) or 150 Mbps (future-proof)

Scenario: 48-port access switch, slow file transfers during peak hours

Assessment:

  • Switch: 48× 1 Gbps ports, 1× 1 Gbps uplink to core
  • Peak uplink utilization: 950 Mbps (95%)
  • Oversubscription: 48:1 (problematic)

Analysis:

Ideal oversubscription: 4:1 to 8:1
Required uplink: 48 Gbps / 6 = 8 Gbps
Available: 10 Gbps uplink port on switch

Solution: Upgrade uplink to 10 Gbps, improves oversubscription to 4.8:1

Example 3: WiFi Capacity for Conference Room

Scenario: 100-person conference room, video-heavy presentations

Assessment:

  • Current: 1× WiFi 5 AP (theoretical 1.3 Gbps, real-world 600 Mbps)
  • Expected load: 75 concurrent clients streaming 1080p video
  • Requirement: 5 Mbps per client for HD streaming

Analysis:

Total required: 75 × 5 Mbps = 375 Mbps
Current capacity: 600 Mbps (sufficient for bandwidth)
BUT: WiFi 5 AP typically supports 30-50 clients efficiently

Client density problem: 75 clients on 1 AP = degraded performance

Solution:

  • Option 1: 2× WiFi 6 APs (each supports 75 clients, provides redundancy)
  • Option 2: 3× WiFi 5 APs (distributed load: 25 clients each)

Recommendation: Option 1 (WiFi 6 for efficiency and future-proofing)

Additional Resources

  • Cisco Network Design Guide: Capacity planning best practices
  • PRTG Bandwidth Calculator: Online bandwidth estimation tool
  • NetFlow Analyzers: Baselining and trending tools

Proper capacity planning prevents performance issues, reduces downtime, and ensures efficient resource utilization. Plan for growth, monitor continuously, and upgrade proactively.

Edit this page